DESCRIPTION (investigator's abstract): Rheumatoid arthritis (RA) is a common, chronic disease that causes substantial morbidity and mortality while utilizing a large portion of health care resources. While recent therapeutic approaches have provided significant improvement in patients' functional status and well being, the selection of specific therapy remains empiric and highly variable. The ability to use surrogate markers to predict which patient is likely to respond to certain anti-rheumatic therapy would offer major advantages in the management of this chronic and progressive disease. Recent evidence has suggested a clear role for proinflammatory cytokines, membrane metalloproteinases, and genetic factors in influencing the severity and progressive nature of the disease. Thus, it is reasonable to suggest that the status of these factors may allow the characterization of patients into 'responder' versus 'non-responder' groups. The intent of this proposal is to utilize material from carefully controlled clinical trials where patient groups have been identified as either responding or not responding to specific therapies. This material will be evaluated for various cytokine, MMP, and genetic factors that may predict response to therapy. The hypothesis is that determining specific patient cytokine profiles, membrane metalloproteinase levels, and/or genetic phenotypes may predict the 'responder/non-responder' status of patients with specific therapy intervention. This hypothesis will be tested by three specific aims which are: Specific Aim #1 - correlate the response of patients with RA to specific therapies with changes in cytokine, matrix metalloproteinases and/or acute phase protein levels. Specific Aim #2 - correlate the response of patients with RA to specific therapies with HLA-DRB1 subtyping and non-MHC gene polymorphism. Specific Aim #3 - develop strategies for initiating prospective clinical trails using predictive cytokine, matrix metalloproteinase, and genetic factors. The unique strength of this proposal is the patient serum and DNA sample bank that is available from several controlled, randomized, blinded, clinical protocols that clearly define patients as responder or non-responder. Retrospective analysis of this information can be utilized to plan prospective clinical trials using potentially predictive patterns of cytokine, matrix metalloproteinases, and genetic factors.